Sunday: 17 December 2006
I had a sneaking suspicion that color vision was complicated, but I really had no idea how complicated. My “Neuroscience” textbook (MF Bear, BW Connors, and MA Paradiso, 1996) is, and I kid you not, an extremely fun and stimulating read, but it only devotes two pages to explaining color vision and not all that well, either. Fortunately we have the wisdom of the internets at hand and there are some marvelous websites devoted to this subject. I’ll present some of those at the end.
I’m limiting this severely, and with the caveat that as in most things I’m no expert. What follows is simply the way I understand it, and I’ve stripped out a vast number of details. Real experts will probably be scandalized.
My goals are to explain the afterimage effects of the post of a couple of days ago and their significance, and in the next post to explain genetic colorblindness, of which there are quite a few forms. So I’ve left out, or simpified almost criminally, a great many things. I’m not going to explain how we see many shades, hues, and luminosities. I’m limiting myself to the four major colors: red, green, yellow, and blue. I’m not going to explain how we see very acutely in dim and in very bright conditions both, over 9 orders of magnitude of light intensity. And I’m not going to do any more than mention that besides the rods and cones of the retina in the eye there are a myriad of other supporting cell types, and each one has its own name.
Basics of light, in five sentences:
I imagine most people have an image of a rainbow, that phenomenon in which white light is broken up into its component colors of blue, green, yellow, and red, and many hues in between. It’s nature’s way of telling us to forget painting class and mixing pigments - the *real* physics is of mixing light. When you mix paints of all colors, you get a muddy gray-black, but mix light of all colors and you get white light. Yellow is considered a primary color in painting class, but the reality is that yellow light is a combination of red and green light. Keep these things in mind, and also remember the order of colors, from short wavelength (blue) through green to yellow to long wavelengths of red.
Basics of rods and cones:
Human eyes have two types of light sensitive cells that pepper the back of the inside of the eye, upon the retina. Images are focused upon the retina and like a digital camera’s CCD the rods and cones are activated by these patterns and colors of light. Rod cells are much more numerous, detect only light intensity and don’t discriminate between colors, and are much more sensitive to light than cone cells. I won’t talk about them further. Cone cells are capable of distinguishing colors of light, but they aren’t very sensitive and are much less numerous except in the very back of the eye’s retina, at the fulvea, where cones are just about all there are.
In most humans, cones come in three types: L-cones respond to long wavelength light, but I’m going to refer to them as Red Cones. M-cones respond to medium wavelength light, but I’m going to call them Green Cones. And S-cones which respond to short wavelength light I will refer to as Blue Cones. Notice: we don’t have Yellow Cones!
Because most of us have three types of cones that establish color perception, normal humans are called trichromats. (However many birds and some reptiles and a very few human women have four types of cones, and are tetrachromats. Turtles have *five* types of cones and are pentachromats!) There are people who have only one or two types of cones, and they are monochromats or dichromats, respectively, and endure one form of color blindness or another, but more about that later.
How cones absorb light:
Cone cells produce large amounts of a pigmented protein called Rhodopsin. The protein is called Opsin, and the pigment is a metabolite of Vitamin A called Retinal. When the protein opsin grabs the pigment and squeezes it, the pigment is able to absorb light of a particular color.
Humans have three genes, at least, for opsin. There’s Opsin OPN1MW, which is made by the green cones and which I will simply call green opsin. Green opsin grabs hold of the retinal pigment and squeezes it in such a way that it absorbs green and yellow light. I expect you to remember that just because I call it “green” doesn’t mean that it only absorbs green - it also absorbs yellow. There’s also Opsin OPN1LW, made by the red cones, and red opsin absorbs red AND yellow light. NOTICE! The green and red cones overlap in the colors of light they absorb! And then there’s OPN1SW, the Blue Opsin that absorbs primarily blue light.
It’s unfortunate that physics makes life complicated - if red cones absorbed *only* red light, and green cones only green, and blue cones only blue, things would be much simpler. But they don’t - retinal is not that discriminating a pigment no matter how cleverly it’s squeezed, and there is overlap. Here’s an absorption spectrum of the three kinds of rhodopsin and the colors they absorb:
The Mystery of Yellow:
We don’t have Yellow Cones!
You can see how this could be a problem. The red cones absorb both red and yellow, and so if that’s all there was to it then yellow objects would look red to you. Even worse, the green cones absorb yellow too, so yellow objects would be displayed twice - as red by red cones, and green by green cones.
Fortunately this doesn’t happen - we do see yellow as yellow and not as red or green, and we see yellow even though we don’t have yellow cones! We are able to do it because cones talk to each other. And how they talk to each other and negotiate who defines what is what makes this color aspect of vision so complicated.
Color Opponents:
In the late 1800s, Ewald Hering noticed a few things. He noticed that there are colors that we don’t perceive in combination, for one thing. We don’t see things as reddish-green, and we don’t see things as bluish-yellow (remember, we’re not talking paints here). He also played around with the afterimages thing I posted on a few days ago. From this website, which is quite good in explaining a lot of interesting things about color vision, here’s another opportunity to follow in Hering’s footsteps:
Remember to stare at the black dot in the center of the colored squares for 30 seconds, then flick your gaze to the white portion on the right to see the afterimages. (As an additional experiment, you can also shut and cover your eyes and see the afterimages that way.) Red always gave a green afterimage, and vice-versa, and blue always gave a yellow afterimage, and vice versa. (More later on why that may not be what we actually see here.)
Hering’s thought was that green and red opposed one another, and yellow and blue opposed one another - they came in pairs, in other words. He hypothesized that there was a cellular, physiological basis for this. What we now know about how our three types of cone cells interact confirms his hypothesis.
Red and Green:
The red-green opponents are probably the easiest to understand. It’s necessary for red and green cone cells to talk to each other and hash it all out because both see yellow. Now there’s an explanation involving center and surround cells, and ON and OFF cells, and I’m not going to get into that. Suffice it to say that red and green cone cells are wired together so that they can compare their information on a particular color they have BOTH received, and do a little subtraction:
If the light the red cone sees is brighter than the same light that the green cone sees, the decision will be to make the output red. And vice-versa, if the light the green cone sees is brighter than the same light the red cone sees, the output will be for green. In this way green cones never output anything unless they see green, and red cones never output anything unless they see red light. Now what about yellow?
Blue and Yellow:
Remember that we have no yellow cone cells but also keep in mind that red and green cones directly detect yellow, as an accident of spectral overlap. The only cone cells left are the blue ones and they detect blue light only. A strong blue light activates them and that is their output. What, then about yellow? A yellow output requires a decision to be made by all three cones, and so they are all wired together. If red and green cones see about equally intense light, and blue cone cells don’t see any light, then the decision is made to output yellow.
Voila! Yellow is rendered only when there’s red and green light, and no blue.
And now for the Afterimages:
Let’s say you stare at a red square. After 30 seconds, your red cones have gotten tired, and run out of oomph. When you look at a white field, or shut your eyes, you’re supposed to see a green afterimage, and that’s because only the green cones are firing and the red ones now are not. Now most of us actually saw a blue afterimage after staring at a red square, and I think that’s probably for one or both of two reasons:
First, the color of the square might not have been suitable, but I don’t think so. I suspect that what’s happening with the pure red square is that with red cones exhausted, and with blue and green cones both firing, the blue simply overwhelmed the green in the afterimage. In other words, it’s a technical detail.
Here’s the test for that. Remember the purple square? I noticed that when you stared at a purple square, the color of the afterimage really WAS green, and Robin confirmed that she saw it too.
I suggest that’s because purple light is a combination of red and blue, and that the red and blue cones have thus been exhausted, leaving only the green cones to fire. With a purple square, we were able to exhaust the blue cones and let the green cones show through.
Now let’s stare at a yellow square, and after 30 seconds, the afterimage is blue. Almost everyone got this one. Remember that your red and green cones are responding to yellow, and after awhile they’ve gotten tired. When you switch to the white field, the only cones capable of acting are the blue cones, and thus you see a blue afterimage.
And what about staring at the blue square? After 30 seconds you’ve tired out your blue cones, which are now registering zero output. When you look at the white square both the red and green cones are stimulated equally, and as we know equal signals from red and green combined with zero from blue produce a yellow impression, and that’s the color of the afterimage. Now there were quite a few folks who saw at best a beige and at worst no afterimage but the white field. How to explain this? Perhaps they have super blue cones that don’t get tired easily!
(And then there’s Karen.)
Putting it all together:
Here’s a very very simple diagram of how it works. Take any teensy patch of retina at the back of the eye, and you’ll see a bunch of cones - most are red and green cones and a minority are blue cones. If you look closely enough you’ll see that the cones in such a tiny patch are wired together. While I’ve represented the wiring as if it’s really wiring, what it really is a dozen or more non-cone cell types that sample outputs from the cones themselves. It’s these cells that actually do the calculations and make the decisions and output them to still more cell types that eventually transmit final decisions to the brain. Nonetheless:
The red cone detects both yellow and red light, and the green cone detects both green and yellow light - their absorption spectra overlap in the yellow region. They communicate and determine who predominates and that is the ultimate output of color information. Red only outputs red and green only outputs green.
The blue cone only outputs blue. It has no overlap problem. And yellow output results when blue cones are quiet but red and green cones are equally stimulated.
Genetics later!
Here are a few webpages that are superlative in their far more detailed and rigorous explanations:
For a very detailed and lengthy tutorial on all aspects of vision.
NIH’s Color Vision Deficiency page.
A less technical page on human color vision.
Another version of the afterimages squares. I see that they don’t use the pure RGB values for a given color, so their red is much closer to my purple.
I’m limiting this severely, and with the caveat that as in most things I’m no expert. What follows is simply the way I understand it, and I’ve stripped out a vast number of details. Real experts will probably be scandalized.
My goals are to explain the afterimage effects of the post of a couple of days ago and their significance, and in the next post to explain genetic colorblindness, of which there are quite a few forms. So I’ve left out, or simpified almost criminally, a great many things. I’m not going to explain how we see many shades, hues, and luminosities. I’m limiting myself to the four major colors: red, green, yellow, and blue. I’m not going to explain how we see very acutely in dim and in very bright conditions both, over 9 orders of magnitude of light intensity. And I’m not going to do any more than mention that besides the rods and cones of the retina in the eye there are a myriad of other supporting cell types, and each one has its own name.
Basics of light, in five sentences:
I imagine most people have an image of a rainbow, that phenomenon in which white light is broken up into its component colors of blue, green, yellow, and red, and many hues in between. It’s nature’s way of telling us to forget painting class and mixing pigments - the *real* physics is of mixing light. When you mix paints of all colors, you get a muddy gray-black, but mix light of all colors and you get white light. Yellow is considered a primary color in painting class, but the reality is that yellow light is a combination of red and green light. Keep these things in mind, and also remember the order of colors, from short wavelength (blue) through green to yellow to long wavelengths of red.
Basics of rods and cones:
Human eyes have two types of light sensitive cells that pepper the back of the inside of the eye, upon the retina. Images are focused upon the retina and like a digital camera’s CCD the rods and cones are activated by these patterns and colors of light. Rod cells are much more numerous, detect only light intensity and don’t discriminate between colors, and are much more sensitive to light than cone cells. I won’t talk about them further. Cone cells are capable of distinguishing colors of light, but they aren’t very sensitive and are much less numerous except in the very back of the eye’s retina, at the fulvea, where cones are just about all there are.
In most humans, cones come in three types: L-cones respond to long wavelength light, but I’m going to refer to them as Red Cones. M-cones respond to medium wavelength light, but I’m going to call them Green Cones. And S-cones which respond to short wavelength light I will refer to as Blue Cones. Notice: we don’t have Yellow Cones!
Because most of us have three types of cones that establish color perception, normal humans are called trichromats. (However many birds and some reptiles and a very few human women have four types of cones, and are tetrachromats. Turtles have *five* types of cones and are pentachromats!) There are people who have only one or two types of cones, and they are monochromats or dichromats, respectively, and endure one form of color blindness or another, but more about that later.
How cones absorb light:
Cone cells produce large amounts of a pigmented protein called Rhodopsin. The protein is called Opsin, and the pigment is a metabolite of Vitamin A called Retinal. When the protein opsin grabs the pigment and squeezes it, the pigment is able to absorb light of a particular color.
Humans have three genes, at least, for opsin. There’s Opsin OPN1MW, which is made by the green cones and which I will simply call green opsin. Green opsin grabs hold of the retinal pigment and squeezes it in such a way that it absorbs green and yellow light. I expect you to remember that just because I call it “green” doesn’t mean that it only absorbs green - it also absorbs yellow. There’s also Opsin OPN1LW, made by the red cones, and red opsin absorbs red AND yellow light. NOTICE! The green and red cones overlap in the colors of light they absorb! And then there’s OPN1SW, the Blue Opsin that absorbs primarily blue light.
It’s unfortunate that physics makes life complicated - if red cones absorbed *only* red light, and green cones only green, and blue cones only blue, things would be much simpler. But they don’t - retinal is not that discriminating a pigment no matter how cleverly it’s squeezed, and there is overlap. Here’s an absorption spectrum of the three kinds of rhodopsin and the colors they absorb:
The Mystery of Yellow:
We don’t have Yellow Cones!
You can see how this could be a problem. The red cones absorb both red and yellow, and so if that’s all there was to it then yellow objects would look red to you. Even worse, the green cones absorb yellow too, so yellow objects would be displayed twice - as red by red cones, and green by green cones.
Fortunately this doesn’t happen - we do see yellow as yellow and not as red or green, and we see yellow even though we don’t have yellow cones! We are able to do it because cones talk to each other. And how they talk to each other and negotiate who defines what is what makes this color aspect of vision so complicated.
Color Opponents:
In the late 1800s, Ewald Hering noticed a few things. He noticed that there are colors that we don’t perceive in combination, for one thing. We don’t see things as reddish-green, and we don’t see things as bluish-yellow (remember, we’re not talking paints here). He also played around with the afterimages thing I posted on a few days ago. From this website, which is quite good in explaining a lot of interesting things about color vision, here’s another opportunity to follow in Hering’s footsteps:
Remember to stare at the black dot in the center of the colored squares for 30 seconds, then flick your gaze to the white portion on the right to see the afterimages. (As an additional experiment, you can also shut and cover your eyes and see the afterimages that way.) Red always gave a green afterimage, and vice-versa, and blue always gave a yellow afterimage, and vice versa. (More later on why that may not be what we actually see here.)
Hering’s thought was that green and red opposed one another, and yellow and blue opposed one another - they came in pairs, in other words. He hypothesized that there was a cellular, physiological basis for this. What we now know about how our three types of cone cells interact confirms his hypothesis.
Red and Green:
The red-green opponents are probably the easiest to understand. It’s necessary for red and green cone cells to talk to each other and hash it all out because both see yellow. Now there’s an explanation involving center and surround cells, and ON and OFF cells, and I’m not going to get into that. Suffice it to say that red and green cone cells are wired together so that they can compare their information on a particular color they have BOTH received, and do a little subtraction:
If the light the red cone sees is brighter than the same light that the green cone sees, the decision will be to make the output red. And vice-versa, if the light the green cone sees is brighter than the same light the red cone sees, the output will be for green. In this way green cones never output anything unless they see green, and red cones never output anything unless they see red light. Now what about yellow?
Blue and Yellow:
Remember that we have no yellow cone cells but also keep in mind that red and green cones directly detect yellow, as an accident of spectral overlap. The only cone cells left are the blue ones and they detect blue light only. A strong blue light activates them and that is their output. What, then about yellow? A yellow output requires a decision to be made by all three cones, and so they are all wired together. If red and green cones see about equally intense light, and blue cone cells don’t see any light, then the decision is made to output yellow.
Voila! Yellow is rendered only when there’s red and green light, and no blue.
And now for the Afterimages:
Let’s say you stare at a red square. After 30 seconds, your red cones have gotten tired, and run out of oomph. When you look at a white field, or shut your eyes, you’re supposed to see a green afterimage, and that’s because only the green cones are firing and the red ones now are not. Now most of us actually saw a blue afterimage after staring at a red square, and I think that’s probably for one or both of two reasons:
First, the color of the square might not have been suitable, but I don’t think so. I suspect that what’s happening with the pure red square is that with red cones exhausted, and with blue and green cones both firing, the blue simply overwhelmed the green in the afterimage. In other words, it’s a technical detail.
Here’s the test for that. Remember the purple square? I noticed that when you stared at a purple square, the color of the afterimage really WAS green, and Robin confirmed that she saw it too.
I suggest that’s because purple light is a combination of red and blue, and that the red and blue cones have thus been exhausted, leaving only the green cones to fire. With a purple square, we were able to exhaust the blue cones and let the green cones show through.
Now let’s stare at a yellow square, and after 30 seconds, the afterimage is blue. Almost everyone got this one. Remember that your red and green cones are responding to yellow, and after awhile they’ve gotten tired. When you switch to the white field, the only cones capable of acting are the blue cones, and thus you see a blue afterimage.
And what about staring at the blue square? After 30 seconds you’ve tired out your blue cones, which are now registering zero output. When you look at the white square both the red and green cones are stimulated equally, and as we know equal signals from red and green combined with zero from blue produce a yellow impression, and that’s the color of the afterimage. Now there were quite a few folks who saw at best a beige and at worst no afterimage but the white field. How to explain this? Perhaps they have super blue cones that don’t get tired easily!
(And then there’s Karen.)
Putting it all together:
Here’s a very very simple diagram of how it works. Take any teensy patch of retina at the back of the eye, and you’ll see a bunch of cones - most are red and green cones and a minority are blue cones. If you look closely enough you’ll see that the cones in such a tiny patch are wired together. While I’ve represented the wiring as if it’s really wiring, what it really is a dozen or more non-cone cell types that sample outputs from the cones themselves. It’s these cells that actually do the calculations and make the decisions and output them to still more cell types that eventually transmit final decisions to the brain. Nonetheless:
The red cone detects both yellow and red light, and the green cone detects both green and yellow light - their absorption spectra overlap in the yellow region. They communicate and determine who predominates and that is the ultimate output of color information. Red only outputs red and green only outputs green.
The blue cone only outputs blue. It has no overlap problem. And yellow output results when blue cones are quiet but red and green cones are equally stimulated.
Genetics later!
Here are a few webpages that are superlative in their far more detailed and rigorous explanations:
For a very detailed and lengthy tutorial on all aspects of vision.
NIH’s Color Vision Deficiency page.
A less technical page on human color vision.
Another version of the afterimages squares. I see that they don’t use the pure RGB values for a given color, so their red is much closer to my purple.
